A friend showed me Siri the other day and after talking to her for a while I had to think, “How could this somehow be applied to the CM realm?” Could we get voice recognition to help us check-in or check-out code? Maybe it could help us check differences or even merge code. OK, I won’t hold my breath on that last one; first we have to get a tool that can merge code seamlessly with little human interaction. That is why a recent story on The New York Times Bits blog drew my interest.
Nick Biltin’s post, “With Apple’s Siri, a Romance Gone Sour,” looks at Siri from what it was supposed to do vs. what it actually does. Nick states that when he asks Siri questions, communication issues occur, including misunderstandings and repetitive responses. He points to a recent study by Gene Munster who put Siri through a series of 1,600 voice tests. Gene used a quiet setting for one half of the test and the streets of Minneapolis for the other half, finding:
- In the quiet setting, Siri understood 89 percent of requests but only accurately answered 68 percent of the time.
- In the street setting, Siri understood his requests 83 percent of the time with a success rate of 62 percent.
Biltin goes on to say that he has now switched over to Google Voice Search, which he finds to be better. However, he does remind users that Siri, according to Apple, is in beta mode and is being improved upon. Regarding Siri’s communication issues, I found a couple of reader comments on Nick’s article to be quite amusing:
Jake from New York City: “I really tried. It ends up being more convenient just doing whatever I'm trying to do myself, whether it's setting an alarm, Googling something, or making a call. Maybe it's voice recognition software in general, but Siri messes it up often enough that it just ends up not being worth using.”
Rex from New Jersey: “I can't understand why Siri will say 'Try again later.' What? Is she busy with something else? Doing her nails?”
From a CM perspective, I can’t do my job if Siri is only correct 62-68 percent of the time. I can’t take the chance with my production baseline with those kinds of numbers. So from a tool perspective, it would be utterly useless to my users as well. Other uses for Siri in the IT arena would be doubtful as well given the high percentage of failure and time criticality.
Maybe Siri is just a novelty item that is a nice toy to play with but with very little value as a viable application for developers, testers, configuration management, and operations users at this time.
Perhaps in the future, a Siri-like tool would be helpful to our profession, but at this time, voice recognition is not where it needs to be for this to be a viable option. It would be awesome to have a tool where you could ask it to backup server A or restore server B—and it could complete the action. It would be nice if developers could simply speak their code out loud and have a Siri-like tool tell them if there are errors as they dictate.
I think these kinds of dreams may be possible, but so far, Siri may not be up to the challenge. Maybe we weren’t ready for Siri, or maybe Siri wasn’t ready for us—and I am afraid to ask her that question.
Joe Townsend has been in the configuration management field for twelve years. He has worked for CNA Life Insurance, RCA, Boeing, UPS, and in state government. Joe has primarily worked with Serena tools, including PVCS Version Manager, Tracker, TeamTrack (Mashups), and Dimensions. He is an administrator for WebFocus and supports Eclipse users.